MicroRNAs (miRNAs) are involved in many diverse biological processes and a single miRNA can modulate the expression of hundreds of gene targets. Identification of miRNA targets is a critical first step for miRNA functional studies. Unfortunately, none of the existing computational tools are sufficient for the accurate prediction of miRNA targets. The major goal of this study is to significantly improve our ability to predict miRNA gene targets by developing and validating a new computational target prediction model. Our preliminary data indicate that machine learning techniques are effective in building improved target prediction models using expression profiling data. Based on these studies, we propose the hypothesis that the application of novel computational methods on large expression profiling datasets will lead to a robust bioinformatics model for miRNA target prediction. Under Specific Aim 1, we will build this model by first performing microarray profiling analyses to identify a large number of miRNA-down regulated genes. The expression profiling data generated from our study will then be used as training data to identify general sequence features that are important to miRNA target prediction. These features will be systematically analyzed in various computational frameworks to build a robust target prediction model, which will later be validated with independent experimental data. Under Specific Aim 2, this new model will be used to develop an online miRNA database for target prediction for all known miRNAs in humans and animals. In addition to bioinformatics target prediction data, the new database will also host data integrated from many heterogeneous sources. Thus, our new database will serve as an integrated web portal to retrieve relevant miRNA functional data with a focus on miRNA target gene regulation. Finally, under Specific Aim 3, we will evaluate the usefulness of the new computational model and database by initiating the development of new miRNA-based therapeutic strategies for disease control. As a specific example, we will test predicted oncogene modulation by natural and artificial miRNAs as a potential new strategy for cancer control. The successful completion of this project will significantly advance our ability to understand the gene regulation by miRNAs in a variety of normal physiological processes and disease states.